Summary of systematic reviews and meta-analyses † addressing the role of genetic and environmental factors on the occurrence of gastric precancerous lesionsCAG – chronic atrophic gastritis; RR – relative risk; VNTR – variable number tandem repeat; IM – intestinal metaplasia; OR – odds ratio; PG – pepsinogen.* the magnitude of the association depends on the method used to assess atrophy; previous studies showed that PG I alone had a low sensitivity for serological definition of chronic atrophic gastritis (Miki, 2006).† systematic reviews and meta-analyses were identified through PubMed search, from its inception to December 2010, under the following expression (gastritis OR chronic OR atroph* OR intestinal metaplasia OR dysplasia) AND (gastric OR stomach) AND (helicobacter pylori OR gene OR polymorphism OR SNPs OR smoking OR tobacco OR cigarette OR salt OR antioxidant OR diet OR lifestyle OR environmental OR behaviour) AND (meta-analysis OR "systematic review").
Gastric carcinoma has been considered an aetiologically heterogeneous entity, with the role of its potential determinants differing with tumour location (Cavaleiro-Pinto et al., 2011; Helicobacter and Cancer Collaborative Group, 2001; Huang et al., 1998; Huang et al., 2003; Ladeiras-Lopes et al., 2008; Larsson et al., 2006; Lunet et al., 2007; Tredaniel et al., 1997; World Cancer Research Fund & American Institute for Cancer Research, 2007) and histological type (Helicobacter and Cancer Collaborative Group, 2001; Huang et al., 1998; Ladeiras-Lopes et al., 2008; Larsson et al., 2006; Lunet et al., 2007; World Cancer Research Fund & American Institute for Cancer Research, 2007). Regarding the latter, Laurén proposed an histo-clinical classification (Laurén, 1965) comprising two main histological types – diffuse and intestinal – with different frequency and distribution across populations (Muñoz & Asvall, 1971; Muñoz & Connelly, 1971). Most gastric carcinomas belong to the intestinal type, representing between 52% and 82% of all gastric cancers (Kaneko & Yoshimura, 2001; Laurén & Nevalainen, 1993; Wu et al., 2009). A higher incidence of intestinal type tumours was observed in males, blacks and older subjects, while the diffuse type had a similar incidence in both genders and was more frequent in younger individuals (Correa et al., 1973; Ekström et al., 2000). Also, there was a wide geographical variation in the frequency of intestinal type tumours, whereas the occurrence of diffuse adenocarcinomas was more uniform across regions (Laurén & Nevalainen, 1993). In addition, the decrease in cancer incidence among migrants from high- to low-risk areas was observed predominantly for tumours of the intestinal type (Correa et al., 1973). These findings were taken as evidence of a relatively greater impact of environmental factors in the aetiology of intestinal type carcinomas, while the diffuse type was considered more dependent on the genetic profile of the individuals (Tahara, 2004). Pelayo Correa (Correa et al., 1975) proposed a model for the development of the intestinal type tumours, according to which the precancerous lesions occur in sequential steps: chronic atrophic gastritis, intestinal metaplasia, and dysplasia. It provided a framework for understanding the role of different environmental and constitutional factors in gastric carcinogenesis, which has evolved with the epidemiologic findings on this topic.
2. Gastric carcinogenesis models
In his initial model (Correa et al., 1975), Correa postulated that both deleterious and protective exposures could modulate the progression towards intestinal type cancers, by acting in different stages of the pathway. With the rediscovery of Helicobacter pylori in 1984 (Marshall & Warren, 1984) and the gradual recognition of its role as a carcinogen, the model was redefined to accommodate the causal relation between H. pylori infection and gastric cancer (Correa, 1992), assuming that its effects were exerted at the early phases of gastric carcinogenesis (Figure 1).
The models proposed by Pelayo Correa more than 15 years ago still provide the essential framework for research on gastric carcinogenesis. The understanding of potentially alternative pathways, the more accurate definition of the endpoints for research, and the identification of the carcinogenesis steps where each of the gastric cancer causal components may act will contribute for a better understanding of cancer aetiology and support the development of preventive strategies.
2.1. The effect of risk factors for gastric cancer across the carcinogenesis pathway
The research on the determinants of gastric cancer precursors has been less extensive than for the cancer endpoints. In Table 1 we present a summary of the systematic reviews and meta-analyses of research conducted to assess the determinants of the lesions that precede the cancer. The contribution of these findings for improvement of the currently accepted gastric carcinogenesis model will be discussed in the following sections of this chapter.
(Adamu et al., 2010)
|14||H. pylori-infected vs. uninfected|
RR=5.0 (95%CI: 3.1-8.3)
Includes studies with follow-up periods ranging from 5 to 32 years
(Peleteiro et al., 2010b)
|PubMed||15||IL1RN VNTR (22 vs. LL)|
CAG and IM:
OR=2.27 (95%CI: 1.40-3.70)
OR=1.65 (95%CI: 1.02-2.66)
OR=2.27 (95%CI: 1.14-4.51)
IL1B-511 (TT vs. CC)
CAG and IM:
OR=1.34 (95%CI: 0.87-2.07)
OR=1.20 (95%CI: 0.70-2.05)
OR=1.94 (95%CI: 1.14-3.31)
TNFA-308 (AA vs. GG)
CAG and IM:
OR=0.93 (95%CI: 0.35-2.43)
(Dias-Neto et al., 2010)
|Salt intake||IM prevalence||PubMed||17||Salted/salty meat intake (highest vs. lowest exposure)|
OR=1.68 (95%CI: 0.98-2.90)
Preference for salted/salty foods or use of table salt (highest vs. lowest exposure)
OR=1.53 (95%CI: 0.72-3.24)
(Weck & Brenner, 2008)
|CAG prevalence evaluated by:|
- gastroscopy with biopsy
- PG I only
- PG I/PG II ratio
- combi-nation of PG I and PG I/PG II ratio
|MEDLINE||66||H. pylori-infected vs. uninfected|
Gastroscopy with biopsy:
OR=6.4 (95%CI: 4.0-10.1)
PG I only *:
OR=0.9 (95%CI: 0.7-1.2)
PG I/PG II ratio:
OR=7.2 (95%CI: 3.1-16.8)
Combination of PG I and PG I/PG II ratio:
OR=5.7 (95%CI: 4.4-7.4)
OR=4.7 (95%CI: 3.7-6.0)
All methods, except PG I only:
OR=6.1 (95%CI: 4.8-7.7)
(Peleteiro et al., 2008)
|IM prevalence||PubMed||22||Systematic review for identification of estimates for IM prevalence among H. pylori-infected subjects in different populations and ecological analysis of the association with smoking|
Pearson correlation coefficient = 0.45 (p=0.02)
(Rokkas et al., 2007)
|MEDLINE||8||H. pylori-eradicated vs. placebo|
CAG in the antrum:
OR=0.554 (95%CI: 0.372-0.825)
CAG in the corpus:
OR=0.209 (95%CI: 0.081-0.538)
IM in the antrum:
OR=0.795 (95%CI: 0.587-1.078)
IM in the corpus:
OR=0.891 (95%CI: 0.633-1.253)
Includes studies with follow-up periods ranging from 10 to 137 months
2.1.1. The role of Helicobacter pylori infection
The association between H. pylori infection and gastric cancer is well established, with several meta-analyses reporting an odds ratio of approximately 2 (Figure 2). The prevalence of H. pylori infection is high (74% in developing and 58% in developed countries, on average) and nearly two-thirds of all gastric cancers occurring worldwide are attributed to it (Parkin, 2006).
Pelayo Correa proposed that H. pylori infection acted at the early phases of the carcinogenesis based on the fact that the inflammatory reaction could disappear after clearance of the bacteria with antibiotic treatment (Correa, 1992), and this has received support from several lines of evidence. On the one hand, the longer the lag between the assessment of H. pylori infection status and the diagnosis of gastric cancer, the stronger is the association between infection and cancer (Helicobacter and Cancer Collaborative Group, 2001; Huang et al., 1998), as H. pylori clearance tends to occur with the progression to the cancer endpoint (Gao et al., 2009; Kokkola et al., 2003). Case-control designs tend to underestimate the relation between infection and gastric cancer, but studies that used more sensitive methods to detect past infection or restricted the analysis to less advanced cases yielded stronger relative risk estimates (Brenner et al., 2004; Ekstrom et al., 2001; Mitchell et al., 2008; Peleteiro et al., 2010a). On the other hand, this is also in accordance with the stronger associations observed between H. pylori and precancerous lesions (Table 1) than with gastric cancer, as depicted in Figure 2.
A direct correlation between H. pylori prevalence and gastric cancer rates is not observed when countries with different patterns of infection and gastric cancer risk are considered, namely because some of them present low gastric cancer incidences despite the high prevalences of infection, the so-called African and Asian enigmas (Holcombe, 1992; Miwa et al., 2002). In the latter settings, the cancer precursor lesions, especially intestinal metaplasia, are also less frequent than expected given the high prevalence of infection (Campbell et al., 2001; Carrilho et al., 2009; Kalebi et al., 2007; Kidd et al., 1999; Oluwasola & Ogunbiyi, 2004), supporting the hypothesis that H. pylori infection acts before their development (in earlier steps of the carcinogenesis) and that other genetic and/or environmental exposures modulate the progression towards cancer (Campbell et al., 2001; Louw et al., 2001; Lunet & Barros, 2003; Mitchell et al., 2002).
220.127.116.11. The impact of Helicobacter pylori eradication
Since the recognition of the causal link between H. pylori and gastric cancer, research has focused on the potential of eradication of the infection as preventive tool. Some clinical trials concluded that H. pylori eradication reduces gastric cancer risk (Fuccio et al., 2009; Ito et al., 2009) (Figure 3). However, one of the trials (Wong et al., 2004) analysed separately the subjects with and without precancerous lesions, and eradication of H. pylori infection was significantly associated with a decreased risk of developing gastric cancer only among the
latter, supporting the hypothesis of the effects of infection in the early stages of gastric carcinogenesis. This is also in accordance with the lack of association between eradication and cancer in a trial conducted in Colombia, which only included subjects with gastric precancerous lesions at baseline evaluation (Correa et al., 2000; Mera et al., 2005).
Also, a meta-analysis from Rokkas et al. (Rokkas et al., 2007) estimated the long-term impact of H. pylori eradication on the incidence of gastric precancerous lesions (Table 1), with significant reduction in chronic atrophic gastritis risk but not for intestinal metaplasia (Figure 3).
Taken together, these results support the irreversibility of intestinal metaplasia, since H. pylori eradication must occur before a point of no return in order to be effective. This represents one more piece of evidence of an early role for H. pylori infection in the gastric carcinogenesis, as it leads to intestinal metaplasia but must be complemented with other factors for progression towards cancer.
2.1.2. The role of lifestyle factors
Although H. pylori infection is the most important gastric cancer determinant, only a small proportion of infected subjects will reach this endpoint (Hsu et al., 2007; Uemura et al., 2001), and several other potential causal components have to be considered.
Stomach cancer is now considered a tobacco-related cancer, with 17% of cases among men and 11% among women being attributed to it in the more developed countries (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2004). Compared to never-smokers, current smokers have a 20% (among women) to 62% (among men) higher risk of gastric cancer, while gastric cancer is 16% (among women) to 34% (among men) more frequent in former smokers (Ladeiras-Lopes et al., 2008). This shows that the interruption of the exposure to tobacco contributes to a reduction in gastric cancer risk, supporting that its effects are exerted at the later steps of carcinogenesis. Furthermore, the risk of cancer was higher in subjects quitting smoking more recently (summary RR for studies with a mean/median follow-up time < 10 vs. ≥ 10 years: 1.39, 95%CI: 1.30-1.49 vs. 1.09, 95%CI: 0.95-1.25, among men) (Ladeiras-Lopes et al., 2008), in accordance with the hypothesis that smoking acts predominantly by promoting the progression from the more advanced precursor lesions to cancer.
Among lifestyle exposures, the relation between smoking and precancerous lesions, especially intestinal metaplasia, has been the more extensively studied, but no systematic reviews of studies quantifying this association are available. The individual reports that have been published yielded relative risk estimates ranging from 1.42 to 4.91 (Kim et al., 2008; Mesquita et al., 2006). An ecological analysis showed a strong correlation between apparent tobacco consumption and the frequency of intestinal metaplasia among H. pylori-infected subjects (Peleteiro et al., 2008), suggesting that the low cigarette consumption observed in developing countries may be a contributory factor for the disruption of the carcinogenesis pathway, precluding the progression to the more advanced lesions. This was also supported by another ecological analysis that showed lower gastric cancer incidence rates in settings with high prevalence of infection and low apparent tobacco consumption than in those where smoking was more frequent (Lunet & Barros, 2003).
Taken together, these results are indicative of a role for smoking in the stages closer to cancer.
The proposed mechanisms by which salt can cause gastric cancer are either the direct damage of the gastric mucosa causing excessive cell replication or an indirect effect by increasing of the mutagenic potential of N-nitroso compounds, which is compatible with the action of salt intake at the initial and late stages of gastric carcinogenesis, respectively (Correa, 1992). More recently, it was shown that the damage caused by salt may also increase gastric H. pylori colonization (Fox et al., 1999; Nozaki et al., 2002), which is also supportive of an early role in the pathway.
Many methodological limitations preclude valid measurements of salt consumption (Chen et al., 1990), and the excretion of sodium in urine over a 24-h period is the method that reflects more accurately the sodium ingested from different sources (World Cancer Research Fund & American Institute for Cancer Research, 2007). The latter, however, was used only in an ecological study assessing the association between salt and intestinal metaplasia (ECP-EURONUT, 1994).
The summary estimates for the relation between total salt use and gastric cancer obtained in the World Cancer Research Fund meta-analysis correspond to a relative risk of approximately 2 (Figure 4). However, only 17 of the 71 studies identified trough systematic review were included in the meta-analysis, due to the large heterogeneity in the presentation of results (World Cancer Research Fund & American Institute for Cancer Research, 2007).
In the meta-analysis conducted by Dias-Neto et al. (Dias-Neto et al., 2010), salted/salty meat intake and preference for salted/salty foods or use of table salt were associated with an approximately 60% increased risk of intestinal metaplasia (Figure 4), but the authors concluded that the large methodological heterogeneity and in the presentation of the results did not allow a more comprehensive quantitative synthesis or a conclusive overall interpretation of the findings.
The evidence currently available on this topic precludes definite conclusions on the magnitude of the effects of salt consumption, overall and in different steps of carcinogenesis.
Based on the inverse association between ingestion of fresh fruits and vegetables and gastric cancer observed in epidemiological studies, antioxidants such as ascorbic acid and beta-carotene were postulated to play a protective role in the stages closer to cancer by acting as free-radical scavengers (Correa, 1992). This protective effect of fruits and vegetables, however, seems to be weaker than initially expected. The summary estimates from several meta-analyses on this topic are closer to 1 when derived from cohort studies, ranging from 0.89 to 0.95 for fruits intake and from 0.89 to 0.98 for vegetables consumption (Figure 5).
However, a stronger protective effect was observed in studies with a follow-up of 10 or more years (OR=0.66, 95%CI: 0.52-0.83 for fruits and OR=0.71, 95%CI: 0.53-0.94 for vegetables in incidence studies) (Lunet et al., 2005), which may be seen as evidence of a relatively early effect of these factors in the gastric carcinogenesis pathway. Research on this topic having cancer precursor lesions as the outcome has seldom been conducted, with few studies reporting conflicting results regarding chronic atrophic gastritis and intestinal metaplasia (Jedrychowski et al., 1999; Kato et al., 2004; Kuwahara et al., 2000; Sierra et al., 2008).
Although it is plausible that naturally occurring antioxidants may inhibit the carcinogenesis progression, there is limited evidence on the specific steps where these factors may act.
18.104.22.168. N-nitroso compounds
N-nitroso compounds were hypothesized to act in gastric carcinogenesis by promoting the synthesis of carcinogens via nitrosation reactions (Correa, 1992). Processed meat is often an important source of exposure to carcinogenic N-nitroso compounds, and a meta-analysis of studies that quantified the association between processed meat consumption and stomach cancer found stronger associations for case-control studies (case-control vs. cohort: OR=1.63, 95%CI: 1.31-2.01 vs. OR=1.24, 95%CI: 0.98-1.56) (Larsson et al., 2006). The relation between these exposures and cancer precursor lesions has seldom been addressed (Sobala et al., 1991; You et al., 1996) and the effect of these compounds in the gastric carcinogenesis pathway remains to be fully understood.
2.1.3. The role of genetic factors
The genetic profile of the individuals was not included in the first versions of the model proposed by Correa (Correa, 1992; Correa et al., 1975) but, along with the advent of new technologies and their use in epidemiological research, several studies addressing the association between genetic polymorphisms and gastric cancer have been conducted. In the aetiological model of gastric cancer, individual genetic susceptibility may be critical in a variety of processes relevant to gastric carcinogenesis, namely mucosal protection, inflammatory response, carcinogen detoxification, antioxidant protection, DNA repair and oncogenes and tumour suppressor genes expression. The most widely studied polymorphisms, and for which more promising results have been achieved, are those related to proinflammatory cytokines, namely within interleukin-1 (IL1) and tumour necrosis factor α (TNFA) gene clusters.
22.214.171.124. Cytokine gene polymorphisms
H. pylori infection induces both interleukin-1β (IL-1β) and tumour necrosis factor-α (TNF-α) production, and these cytokines inhibit gastric acid secretion, leading to the development of gastric precancerous lesions and cancer (El-Omar et al., 2000; Hwang et al., 2002). The IL1B gene codes for the IL-1β and the IL1RN gene for an anti-inflammatory cytokine, interleukin-1 receptor antagonist (IL-1ra). Polymorphisms within the IL1B gene increase IL-1β expression and IL-1ra binds to the IL-1 receptors, modulating the pro-inflammatory effects of IL-1β. Regarding the IL1RN gene, a variable number tandem repeat (VNTR) polymorphism has been detected within intron 2, and five allelic variants have been identified in the number of repeats varying from 2 to 6 (El-Omar, 2001; Gonzalez et al., 2002). The ability of H. pylori to infect and remain in the human stomach induces a chronic inflammatory response, which may be of variable magnitude depending on the genetic make-up of the host. Most of the single nucleotide polymorphisms (SNPs) studied are situated in the gene promoter region and play important roles in modulating gene expression and thus the inflammatory response.
Previous meta-analyses have shown an increased gastric cancer risk associated with polymorphisms in IL1RN, IL1B-511 and TNFA-308 (Figure 6). The IL1RN*22 genotype increases the risk of gastric precancerous lesions, suggesting a role for this polymorphism in the early stages of gastric carcinogenesis, while positive associations between IL1B-511 TT genotype and gastric precancerous lesions only became apparent when studies addressing intestinal metaplasia as the outcome were considered (Figure 6). These associations were
stronger among studies conducted in samples with high H. pylori prevalence, in accordance to what is known about the gene function and its potential interaction with infection. No overall association was found for TNFA-308 AA genotype (Figure 6).
2.2. Other markers of gastric cancer development
The currently available evidence shows no substantial aetiological differences between the main Laurén subtypes, despite the relatively small number of studies addressing the effect of environmental exposures on the risk of gastric cancer according to histological subtypes (Figures 2 and 5). This may partially reflect misclassification of the histological type, due to inter-observer variability, the type of specimen available for diagnosis, and the proportion of tumours classified as unknown (Carneiro et al., 2007).
An additional concern is the ability for the classification proposed by Laurén to define aetiologically homogeneous subgroups of gastric cancer cases. The cascade of events that involve intestinal differentiation is mediated by CDX1 and/or CDX2 (Guo et al., 2004), and may result in the development of both intestinal and diffuse gastric carcinoma (Almeida et al., 2003). In particular, CDX2 expression is regarded as a marker of the intestinal epithelial phenotype, and the transdifferentiation of normal epithelia has been experimentally induced by changes in local environment (Marchetti et al., 2003), which supports the hypothesis that environmental exposures may modulate the CDX2 expression. This may be seen as an early marker of intestinal differentiation, that may be used as an endpoint occurring in the gastric carcinogenesis pathway even earlier than chronic atrophic gastritis or intestinal metaplasia. Research relying on these tools to define the outcomes is still scarce (Yuasa et al., 2009; Yuasa et al., 2005), but may be important to understand the aetiological heterogeneity of gastric cancer.
Histopathological and histochemical studies allowed the identification of two main types of intestinal metaplasia. The complete, also designated type I, and the incomplete, comprising types II and III (Filipe & Jass, 1986). In the classical multistep model of the gastric precancerous process, incomplete follows complete intestinal metaplasia sequentially (Correa, 1992). However, according to the patterns of mucin expression observed within each intestinal metaplasia type, it has been hypothesised that the complete and incomplete types of intestinal metaplasia may represent two alternative pathways, rather than successive steps; or that type II may represent a first step in the pathway, which may evolve to type I or to type III (Reis et al., 1999). The evaluation of specific risk factors for these endpoints may clarify the gastric carcinogenesis pathways and the role of environmental exposures in the aetiology of cancer (Peleteiro et al., 2007; Pintalhao et al., 2010).
The accumulated evidence so far led to the gradual acceptance and better understanding of the role of H. pylori infection and smoking in gastric carcinogenesis. For other exposures, however, there is much less robust evidence on the magnitude of the associations or their role throughout carcinogenesis. This allows an update of the model proposed by Correa, that still provides the best framework for gastric cancer etiological research, taking into account the evidence generated in the last two decades (Figure 7).
Research relying on more accurate tools to define specific gastric cancer subtypes and the evaluation of specific risk factors for early endpoints in the gastric carcinogenesis pathway may further contribute to the understanding of gastric cancer aetiology.